Multi-fabric physical simulation 3D garment virtual fitting method and system and medium

Abstract

The present application relates to the technical field of virtual fitting, in particular to a multi-fabric physical simulation 3D garment virtual fitting method, system and medium, comprising the following steps: collecting and standardizing fabric characteristic data, storing calibration data and texture data in association according to fabric identification, constructing a fabric parameter database, reading target fabric data and establishing initial texture mapping, iteratively calibrating simulation parameters according to deformation deviation, determining target convergence conditions and solving configurations in combination with posture changes, executing physical simulation, and performing local recalculation or texture updating and three-dimensional rendering according to convergence judgment. The present application stores data in association and calibrates simulation parameters, so that different fabrics obtain simulation parameters corresponding to actual deformation states, and through posture-driven solving and regional feedback calculation, the changes in garment form and texture are kept associated with target fabrics and human postures, thereby reducing the deviation between virtual fitting states and actual wearing states.

Description

Technical Field

[0001] This invention relates to the field of virtual fitting technology, and in particular to a method, system and medium for virtual fitting of 3D clothing using multi-fabric physical simulation. Background Technology

[0002] The field of virtual try-on technology involves human digital models, clothing digital models, clothing pattern data, fabric attribute data, graphics rendering, collision detection, and clothing shape calculation. It is primarily used to represent the relationship between people and clothing in a computer environment and is widely applied in industries such as clothing design, clothing customization, online retail, digital display, and education and training. Traditional 3D virtual try-on refers to a method that uses human body dimensions, postures, and clothing structure data, employing techniques such as human body modeling, clothing modeling, pattern stitching, mesh processing, and graphics rendering to calculate and display the position, shape, folds, and drape of clothing on a human body model.

[0003] Traditional 3D virtual try-on for clothing builds models based on human body size data, posture data, and clothing structure data, and performs mesh calculations and graphic rendering. Fabric attributes are mostly described using fixed parameters or a few general parameters. It is difficult to accurately distinguish the differences in mechanical response and deformation between different types of fabrics. When the human body posture changes, the clothing state is still calculated according to uniform solution conditions. There is a lack of synchronous correlation between fabric texture and mesh deformation, resulting in deviations between stretching, draping, wrinkles, and dynamic changes and the actual wearing state. This limits the adaptability and image realism of virtual try-on for multiple types of clothing. Summary of the Invention

[0004] The purpose of this invention is to provide a method, system, and medium for virtual try-on of 3D clothing using multi-fabric physical simulation. By associating and organizing characteristic data, calibration data, and texture data of different fabrics, the simulation parameters of the target fabric are calibrated. Furthermore, the solution configuration for the physical simulation is determined by combining changes in human posture. This allows clothing made of different fabrics to undergo deformation calculations, local feedback corrections, and texture updates according to their corresponding fabric properties. This addresses the problem that when different types of fabrics are simulated using uniform physical parameters, the changes in clothing shape, wrinkles, drape, and texture do not correspond to the actual fabric state. To achieve the above objectives, the present invention provides a method for virtual try-on of 3D clothing using multi-fabric physical simulation, comprising the following steps: S1: Collect fabric characteristic data, calibration data and texture data, standardize the fabric characteristic data, and store the fabric characteristic data, calibration data and texture data according to the fabric identifier to build a fabric parameter database; S2: Obtain the clothing mesh, target fabric identifier, human body model and pose sequence; read the target fabric data from the fabric parameter database according to the target fabric identifier; extract the target texture data from the target fabric data; and establish an initial texture mapping between the target texture data and the clothing mesh. S3: Generate current simulation parameters based on the target fabric data, perform calibration simulation based on the calibration data and calculate deformation deviation. When the deformation deviation exceeds the calibration range, update the current simulation parameters and repeat the calibration simulation until the deformation deviation enters the calibration range, and determine the current simulation parameters as the target simulation parameters. S4: Drive the human body model according to the posture sequence, determine the target convergence condition and solution configuration according to the changes in adjacent posture frames and the target simulation parameters, and perform physical simulation based on the human body model, the clothing mesh, the target simulation parameters and the solution configuration to obtain the clothing simulation state; S5: When the clothing simulation state does not meet the target convergence condition, locate the deviation region according to the state deviation, update the solution configuration corresponding to the deviation region, and perform local physical simulation and then re-judge. When the clothing simulation state meets the target convergence condition, update the initial texture mapping according to the clothing simulation state and perform three-dimensional rendering to output a three-dimensional fitting screen.

[0005] Furthermore, the fabric characteristic data is standardized into standardized fabric parameters after being standardized in terms of dimensions and verified for data validity; each data record in the fabric parameter database is indexed by the fabric identifier, and the standardized fabric parameters, the calibration data, and the texture data are associated and written into the same data record.

[0006] Further, data records indexed by the target fabric identifier are retrieved according to the target fabric identifier, and the retrieved data records are determined as the target fabric data; based on the topological relationship between the grid vertices and grid patches in the garment grid, a texture coordinate set corresponding to the grid vertex is generated, the arrangement direction of the texture coordinate set is corrected according to the local direction of the grid patch, and the target texture data is mapped to the garment grid according to the corrected texture coordinate set to form the initial texture mapping.

[0007] Further, target calibration data is extracted from the target fabric data, the target calibration data representing the measured deformation state of the target fabric corresponding to the target fabric data under boundary conditions; based on the clothing mesh and the current simulation parameters, the calibration simulation is performed according to the boundary conditions to obtain the virtual deformation state; the deformation deviation is generated by comparing the corresponding deformation state components in the virtual deformation state and the measured deformation state.

[0008] Furthermore, when any of the deformation state components in the deformation deviation exceeds the calibration range, the deviation type is determined based on the deformation state component that exceeds the calibration range, the parameter item corresponding to the deviation type in the current simulation parameters is updated, and the calibration simulation is re-executed according to the boundary conditions; when all the deformation state components enter the calibration range, the calibration simulation is terminated, and the current simulation parameters are determined as the target simulation parameters.

[0009] Further, the posture change state is determined based on the human joint motion and human surface displacement corresponding to adjacent posture frames in the posture sequence; the calculation step size and solution round of the clothing mesh are determined based on the degree of posture change represented by the posture change state and the fabric physical properties represented by the target simulation parameters, forming the solution configuration; the target convergence condition is determined based on the deformation constraints defined by the target simulation parameters.

[0010] Further, the constraint residuals of each grid region in the clothing simulation state are calculated, and the state deviation is determined based on the difference between the constraint residuals and the convergence range defined by the target convergence condition. The grid region where the constraint residuals exceed the convergence range is defined as the deviation region. The solver configuration is adjusted according to the state deviation corresponding to the deviation region. Based on the adjusted solver configuration, the local physical simulation is performed on the deviation region to update the clothing simulation state, and the clothing simulation state is re-evaluated to determine whether it meets the target convergence condition.

[0011] Furthermore, when the clothing simulation state satisfies the target convergence condition, the local deformation relationship is determined based on the change in vertex position of the mesh facets in the clothing mesh before and after the physical simulation, and the initial texture mapping is updated according to the local deformation relationship to form an updated texture mapping; the three-dimensional rendering is performed based on the human body model, the clothing mesh processed by the physical simulation, and the updated texture mapping to output the three-dimensional fitting screen.

[0012] On the other hand, a multi-fabric physical simulation 3D clothing virtual try-on system is provided. This system is applied to the multi-fabric physical simulation 3D clothing virtual try-on method, including: The parameter database module collects fabric characteristic data, calibration data, and texture data, standardizes the fabric characteristic data, and stores the fabric characteristic data, calibration data, and texture data according to the fabric identifier to construct a fabric parameter database. The data mapping module acquires clothing mesh, target fabric identifier, human body model and pose sequence, reads target fabric data from the fabric parameter database according to the target fabric identifier, extracts target texture data from the target fabric data, and establishes an initial texture mapping between the target texture data and the clothing mesh. The parameter calibration module generates current simulation parameters based on the target fabric data, performs calibration simulation and calculates deformation deviation based on the calibration data, and updates the current simulation parameters and repeats the calibration simulation when the deformation deviation exceeds the calibration range until the deformation deviation enters the calibration range, and then determines the current simulation parameters as the target simulation parameters. The physical simulation module drives the human body model based on the posture sequence, determines the target convergence condition and solution configuration based on the changes in adjacent posture frames and the target simulation parameters, and performs physical simulation based on the human body model, the clothing mesh, the target simulation parameters and the solution configuration to obtain the clothing simulation state; The feedback rendering module, when the clothing simulation state does not meet the target convergence condition, locates the deviation region according to the state deviation, updates the solution configuration corresponding to the deviation region, and performs local physical simulation and then re-judges. When the clothing simulation state meets the target convergence condition, it updates the initial texture mapping according to the clothing simulation state and performs 3D rendering to output a 3D fitting screen.

[0013] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a multi-fabric physical simulation 3D clothing virtual try-on system. The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: By standardizing fabric characteristic data and storing it in association with calibration and texture data according to fabric identifiers, a parameter calling basis for different fabrics is formed, enabling the target fabric to obtain corresponding simulation input and texture data. Through calibration simulation, deformation deviations are compared and the current simulation parameters are updated cyclically, establishing a correspondence between the target simulation parameters and the measured deformation state of the fabric. Convergence conditions and solution configurations are determined based on posture changes and target simulation parameters, allowing human posture-driven processes and fabric differences to jointly participate in clothing mesh calculation. When the clothing simulation state does not meet the convergence conditions, the deviation region is located and local physical simulation is performed; when the conditions are met, the texture mapping is updated and rendered, ensuring that clothing shape and texture changes remain correlated with the target fabric and human posture. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart of the method of the present invention; Figure 2 A data flow diagram was constructed for the fabric parameter database of this invention; Figure 3 This is a diagram showing the mapping relationship between the target texture data and the clothing mesh in this invention; Figure 4 This is the closed-loop diagram of the simulation parameter calibration feedback of the present invention; Figure 5 This is the state transition diagram for attitude-driven physical simulation and feedback rendering of the present invention; Figure 6 This is a data interaction diagram of the system modules of the present invention. Detailed Implementation

[0016] The technical solutions described in this invention will be further explained below with reference to the embodiments and accompanying drawings. The following embodiments take the multi-fabric physical simulation 3D clothing virtual try-on method as the main extension object, and the corresponding system and computer-readable storage medium will be described in connection with the multi-fabric physical simulation 3D clothing virtual try-on method.

[0017] The fabric characteristic data, calibration data, texture data, standardized fabric parameters, target fabric data, target texture data, current simulation parameters, target simulation parameters, garment simulation state, state deviation, deviation region, initial texture mapping, and updated texture mapping used in the embodiments all maintain the same technical meaning as the technical solutions described in the application documents. The fabric characteristic data is used to characterize the properties required for the fabric to participate in physical simulation, the calibration data is used to characterize the measured deformation state of the target fabric under corresponding boundary conditions, and the texture data is used to form the visual texture of the garment mesh surface. Example

[0018] Please see Figures 1 to 5This embodiment provides a multi-fabric physical simulation 3D virtual try-on method for clothing. In this embodiment, a fabric parameter database serves as a unified association carrier for multiple types of fabric data, a target fabric identifier serves as the target data retrieval entry point, target simulation parameters are generated through calibration simulation, and physical simulation is driven by posture sequences to model the human body and clothing mesh. When the clothing simulation state does not meet the target convergence condition, local physical simulation is performed on the deviation area. When the clothing simulation state meets the target convergence condition, the texture mapping is updated and a 3D try-on image is output. The multi-fabric physical simulation 3D virtual try-on method forms a continuous processing chain from fabric data acquisition, target data reading, simulation parameter calibration, posture-driven solution, local feedback correction to 3D rendering output.

[0019] S1: Collect fabric characteristic data, calibration data, and texture data; standardize the fabric characteristic data; associate and store the fabric characteristic data, calibration data, and texture data according to the fabric identifier; and construct a fabric parameter database.

[0020] S11: Obtain the fabric characteristic data, calibration data, and texture data corresponding to the same fabric identifier. The fabric identifier serves as the association key for related data of the same category of fabrics, enabling subsequent processing to retrieve the data record corresponding to the target fabric based on a single identifier. Before the collected data enters the standardization process, maintain the correspondence between the fabric identifier and each data item to avoid data from different fabrics being mixed up during storage.

[0021] S12: Perform dimensional unification and data validity verification on the fabric characteristic data to form standardized fabric parameters. Dimensional unification is used to convert fabric characteristic data that can participate in the same simulation parameter generation process to a consistent data expression standard. Data validity verification is used to confirm that the fabric characteristic data to be written into the data record can be read and participate in subsequent processing. Fabric characteristic data that has completed dimensional unification and data validity verification is determined as standardized fabric parameters. Data that has not formed standardized fabric parameters will not enter the subsequent simulation parameter generation process. The verification status of this step is that the standardized fabric parameters can maintain association with the corresponding fabric identifier and can be read by the fabric parameter database.

[0022] S13: Establish a data record indexed by the fabric identifier in the fabric parameter database, and write standardized fabric parameters, calibration data, and texture data into the same data record. After writing, the same data record simultaneously provides the standardized fabric parameters required for subsequent parameter generation, the calibration data required for calibration simulation, and the texture data required for texture mapping. The fabric parameter database does not output separate data sets, but rather associated data records located by the fabric identifier, thus providing a definite data source for reading the target fabric data in S2.

[0023] S2: Obtain the clothing mesh, target fabric identifier, human body model and pose sequence, read the target fabric data from the fabric parameter database according to the target fabric identifier, extract the target texture data from the target fabric data, and establish the initial texture mapping between the target texture data and the clothing mesh.

[0024] S21: Obtain the clothing mesh, target fabric identifier, human body model, and pose sequence. The clothing mesh serves as the object for calculating clothing shape and texture; the human body model serves as the object for calculating pose and clothing wearing relationships; the pose sequence is used to represent the changes in human body state between consecutive pose frames; and the target fabric identifier is used to determine the target fabric data used in this virtual try-on from the fabric parameter database. When these input objects enter subsequent processing, they maintain the corresponding relationship under the same try-on task.

[0025] S22: Retrieve data records indexed by the target fabric identifier, and identify the retrieved data records as the target fabric data. The target fabric data includes standardized fabric parameters, calibration data, and texture data associated with the target fabric identifier. Once a data record is identified as the target fabric data, it provides corresponding data items to the parameter generation process, calibration simulation process, and texture mapping process, respectively. The verifiable output of this process is to obtain a data record with a consistent fabric attribution relationship based on the target fabric identifier.

[0026] S23: Extract target texture data from the target fabric data. Generate a set of texture coordinates corresponding to the mesh vertices based on the topological relationship between mesh vertices and mesh faces in the garment mesh. The texture coordinate set is used to establish the corresponding position of the target texture data on the surface of the garment mesh, and the topological relationship between mesh vertices and mesh faces is used to constrain the continuous correspondence of the texture coordinate set between adjacent mesh faces. Correct the arrangement direction of the texture coordinate set according to the local orientation of the mesh face, so that the texture coordinate set corresponds to the local orientation of the mesh face. Then, map the target texture data to the garment mesh according to the corrected texture coordinate set to form the initial texture mapping. The initial texture mapping serves as the input for texture update in S5, preserving the initial correspondence between the target texture data and the garment mesh.

[0027] S3: Generate current simulation parameters based on target fabric data, perform calibration simulation based on calibration data and calculate deformation deviation. When the deformation deviation exceeds the calibration range, update the current simulation parameters and repeat the calibration simulation until the deformation deviation enters the calibration range, and then determine the current simulation parameters as the target simulation parameters.

[0028] S31: Obtain standardized fabric parameters from the target fabric data to generate simulation parameters, and generate the current simulation parameters based on the standardized fabric parameters. The current simulation parameters are the parameter state for entering this round of calibration simulation, and are used to control the deformation calculation of the garment mesh under calibration boundary conditions.

[0029] S32: Extract target calibration data from the target fabric data. Target calibration data characterizes the measured deformation state of the target fabric corresponding to the target fabric data under boundary conditions. Boundary conditions define the basis for comparison between the calibration simulation and the measured deformation state. Based on the garment mesh and current simulation parameters, perform a calibration simulation according to the boundary conditions to obtain a virtual deformation state. The virtual deformation state and the measured deformation state are compared using corresponding deformation state components, ensuring that the compared objects have consistent fabric attribution and boundary conditions.

[0030] S33: Compare the corresponding deformation state components in the virtual deformation state and the measured deformation state to generate a deformation deviation. The deformation deviation is used to characterize the difference between the virtual deformation state under the current simulation parameters and the measured deformation state characterized by the target calibration data. The deformation deviation is only generated between corresponding deformation state components to avoid treating non-corresponding state quantities as the same comparison object.

[0031] S34: Determine if the deformation deviation is within the calibration range. When any deformation state component in the deformation deviation exceeds the calibration range, determine the deviation type based on the deformation state component that exceeds the calibration range, and then update the parameter item corresponding to the deviation type in the current simulation parameters. The updated current simulation parameters serve as the input for the next round of calibration simulation, and the virtual deformation state is re-obtained and the deformation deviation is regenerated under the same boundary conditions. Thus, each round of updates is based on the deformation deviation and deviation type of the current round, with the updated current simulation parameters as the cyclic input, and the loop terminates when all deformation state components are within the calibration range.

[0032] S35: When all deformation state components are within the calibration range, terminate the calibration simulation and determine the current simulation parameters as the target simulation parameters. The target simulation parameters retain the parameter states that have passed the target calibration data verification and are used as the input of the fabric physical parameters for the S4 physical simulation. The verification method for the calibration process is that the deformation deviation enters the calibration range. The subsequent uses of the target simulation parameters are to participate in the formation of the target convergence conditions, solution configuration, and clothing simulation state.

[0033] S4: Drive the human body model based on the posture sequence, determine the target convergence condition and solution configuration based on the changes in adjacent posture frames and target simulation parameters, and perform physical simulation based on the human body model, clothing mesh, target simulation parameters and solution configuration to obtain the clothing simulation state.

[0034] S41: The human body model is driven according to the sequential relationship of the posture frames in the posture sequence, so that the human body model forms a continuous posture state corresponding to the posture sequence. The changes between adjacent posture frames serve as the basis for determining the solution configuration, and the state of the human body model in the current posture frame serves as the human body action boundary for the physical simulation of the clothing mesh. The posture sequence does not directly generate a 3D fitting image, but first drives the human body model to form a continuous posture state, and then the human body model, clothing mesh, and target simulation parameters jointly enter the physical simulation.

[0035] S42: Determine the posture change state based on the human joint motion and human surface displacement corresponding to adjacent posture frames in the posture sequence. Both human joint motion and human surface displacement originate from changes in adjacent posture frames; the posture change state characterizes the degree of posture change between current adjacent posture frames. The posture change state is only used to determine the solution configuration and is not output as the final fitting result.

[0036] S43: Based on the degree of posture change represented by the posture change state and the physical properties of the fabric represented by the target simulation parameters, determine the calculation step size and solution rounds of the clothing mesh to form a solution configuration. The posture change state provides the dynamic basis for changes in adjacent posture frames, the target simulation parameters provide the simulation basis for the target fabric, and the calculation step size and solution rounds together define the solution method for the physical simulation of the clothing mesh in this round. After the solution configuration is formed, it enters the physical simulation process together with the human body model, clothing mesh, and target simulation parameters.

[0037] S44: Determine the target convergence condition based on the deformation constraints defined by the target simulation parameters. The target convergence condition is used to determine whether the clothing simulation state has reached the convergence state corresponding to the target simulation parameters, and provides a decision boundary for the deviation region location and local physical simulation in S5. The target convergence condition and the solution configuration play different roles; the target convergence condition is used for convergence judgment, while the solution configuration is used to control the solution process of the physical simulation.

[0038] S45: Performs physical simulation based on the human body model, clothing mesh, target simulation parameters, and solver configuration. The human body model provides the current posture state, the clothing mesh provides the object to be solved, the target simulation parameters provide the simulation parameters corresponding to the target fabric, and the solver configuration provides the calculation step size and solution rounds. After the physical simulation is completed, a clothing simulation state is formed. The clothing simulation state represents the deformation state of the clothing mesh under the current human posture and target fabric parameters, and is passed to S5 for convergence judgment.

[0039] S5: When the clothing simulation state does not meet the target convergence condition, locate the deviation area according to the state deviation, update the solution configuration corresponding to the deviation area, and re-judge after performing local physical simulation. When the clothing simulation state meets the target convergence condition, update the initial texture mapping according to the clothing simulation state and perform 3D rendering to output the 3D fitting screen.

[0040] S51: Calculate the constraint residuals for each grid region in the clothing simulation state. The constraint residuals characterize the difference between the current state of each grid region and the convergence range defined by the target convergence condition. The state deviation is determined based on the difference between the constraint residuals and the convergence range, and it is determined whether the constraint residuals of each grid region are within the convergence range. Grid regions whose constraint residuals are within the convergence range are not identified as deviation regions, while grid regions whose constraint residuals exceed the convergence range are identified as deviation regions. Therefore, deviation regions have a definite source from the clothing simulation state, and state deviations have a definite source from the comparison result of the constraint residuals and the convergence range.

[0041] S52: When a deviation region exists, adjust the solver configuration corresponding to the deviation region based on the state deviation. The adjusted solver configuration is only used for subsequent local physical simulations of the deviation region; mesh regions not identified as deviation regions retain their current clothing simulation state. Perform local physical simulations on the deviation region based on the adjusted solver configuration, and update the clothing simulation state with the results of the local physical simulation. The updated clothing simulation state re-enters S51, recalculates the constraint residuals, and determines whether the target convergence condition is met. The loop trigger condition for the feedback process is that the clothing simulation state does not meet the target convergence condition; each update is based on the state deviation corresponding to the deviation region; the loop output is the updated clothing simulation state; and the termination condition is that the clothing simulation state meets the target convergence condition.

[0042] S53: When the clothing simulation state meets the target convergence condition, the local deformation relationship is determined based on the changes in the vertex positions of the mesh faces before and after the physical simulation. The local deformation relationship is used to characterize the local changes of the mesh faces relative to the initial state after the physical simulation, and serves as the basis for updating the initial texture mapping. The initial texture mapping is updated according to the local deformation relationship, so that the corresponding positions of the target texture data on the surface of the clothing mesh are adjusted with the local deformation of the clothing mesh, forming the updated texture mapping.

[0043] S54: Based on the human body model, the physically simulated clothing mesh, and the updated texture map, 3D rendering is performed to output a 3D virtual try-on screen. The human body model is used to present the human posture driven by the posture sequence, the physically simulated clothing mesh is used to present the shape of the target fabric clothing on the human body model, and the updated texture map is used to present the target texture data adjusted according to the local deformation of the mesh facets. The 3D virtual try-on screen, as the final output of the multi-fabric physically simulated 3D clothing virtual try-on method, can verify whether the aforementioned processing chain is complete through the consistent correspondence between the clothing mesh shape, the human body model posture, and the texture position.

[0044] Through steps S1 to S5, the fabric parameter database provides a source of related data for reading the target fabric data. The target fabric data provides input for texture mapping and simulation parameter calibration. The target simulation parameters and posture change state jointly participate in solving the configuration and determining the target convergence condition. The clothing simulation state is compared with the constraint residuals to form deviation regions and state deviations. The deviation regions are updated with local physical simulation feedback to update the clothing simulation state. After the target convergence condition is met, the texture mapping is updated and a 3D fitting image is output. Each intermediate data has a generation position and a subsequent receiving position, forming a continuous data flow and feedback closed loop. Example

[0045] Please see Figure 1 and Figure 6 This embodiment provides a multi-fabric physical simulation 3D clothing virtual try-on system. The difference between this embodiment and Embodiment 1 is that this embodiment uses a modular collaborative approach to carry out the processing chain of the method in Embodiment 1. The parameter library module, data mapping module, parameter calibration module, physical simulation module, and feedback rendering module respectively receive the data or status generated by the preceding modules and pass the processing results to the subsequent modules, finally outputting a three-dimensional try-on screen.

[0046] The parameter database construction module receives fabric characteristic data, calibration data, and texture data. It standardizes the fabric characteristic data and associates and stores the fabric characteristic data, calibration data, and texture data in the fabric parameter database according to the fabric identifier. The parameter database construction module outputs the fabric parameter database, and the associated data records in the fabric parameter database are read by the data mapping module according to the target fabric identifier.

[0047] The data mapping module receives the clothing mesh, target fabric identifier, human body model, and pose sequence. It reads the target fabric data from the fabric parameter database according to the target fabric identifier, extracts the target texture data from the target fabric data, and establishes an initial texture mapping between the target texture data and the clothing mesh. The data mapping module then provides the target fabric data to the parameter calibration module and the human body model, pose sequence, clothing mesh, and initial texture mapping to the physical simulation module or the feedback rendering module.

[0048] The parameter calibration module generates current simulation parameters based on the target fabric data, performs calibration simulation based on the calibration data, and calculates deformation deviation. When the deformation deviation exceeds the calibration range, the parameter calibration module updates the current simulation parameters and repeats the calibration simulation; when the deformation deviation enters the calibration range, the parameter calibration module determines the current simulation parameters as the target simulation parameters. The target simulation parameters are sent to the physical simulation module to determine the target convergence conditions and solution configuration, and participate in the physical simulation of the garment mesh.

[0049] The physics simulation module drives the human body model based on the posture sequence, determines the target convergence conditions and solution configuration based on the changes in adjacent posture frames and the target simulation parameters, and performs physics simulation based on the human body model, clothing mesh, target simulation parameters, and solution configuration to obtain the clothing simulation state. The physics simulation module provides the clothing simulation state, target convergence conditions, and solution configuration to the feedback rendering module.

[0050] The feedback rendering module receives the clothing simulation state, the target convergence condition, the solver configuration, and the initial texture mapping. When the clothing simulation state does not meet the target convergence condition, the feedback rendering module locates the deviation region based on the state deviation, updates the solver configuration corresponding to the deviation region, performs local physical simulation and updates the clothing simulation state, and then returns the updated clothing simulation state to the convergence judgment process. When the clothing simulation state meets the target convergence condition, the feedback rendering module updates the initial texture mapping based on the clothing simulation state and performs 3D rendering, outputting a 3D fitting screen. Example

[0051] Please see Figure 1 and Figure 6 This embodiment provides a computer-readable storage medium. The difference between this embodiment and Embodiment 1 is that this embodiment uses a computer program to carry out and implement the processing functions of the aforementioned system. When the computer program is executed by the processor, the processing logic corresponding to the parameter library module, data mapping module, parameter calibration module, physical simulation module, and feedback rendering module is executed.

[0052] After the computer program stored in the computer-readable storage medium is invoked by the processor, it first executes the process of standardizing fabric characteristic data and constructing a fabric parameter database, then executes the process of reading target fabric data and establishing initial texture mapping, followed by the process of generating current simulation parameters, calibrating simulation, comparing deformation deviations, and determining target simulation parameters. Afterwards, it executes the process of attitude sequence driving, solving configuration determination, determining target convergence conditions, and physical simulation of the clothing mesh, and finally executes the process of deviation region localization, local physical simulation, texture mapping update, and 3D rendering. The execution order of the computer program corresponds to S1 to S5 in Example 1, and the intermediate data and output results generated by the execution of the computer program are consistent with those in Example 1 and Example 2.

[0053] When the deformation deviation exceeds the calibration range, the computer program initiates an update and recalibration process for the current simulation parameters; when the deformation deviation falls within the calibration range, the computer program determines the current simulation parameters as the target simulation parameters. When the clothing simulation state does not meet the target convergence condition, the computer program updates the solution configuration corresponding to the deviation region and performs local physical simulation; when the clothing simulation state meets the target convergence condition, the computer program updates the initial texture mapping and performs 3D rendering. Thus, the program execution relationship carried by the computer-readable storage medium can correspondingly realize the data processing, judgment, looping, and feedback processes of the aforementioned system.

[0054] The above embodiments are used to illustrate the data objects, processing steps, judgment conditions, loop termination conditions, module collaboration relationships, and carrier inheritance relationships in the technical solutions described in this invention. The data field organization, processing order, parameter sources, module division, and program carrying form described in the embodiments are used to explain the possible implementations of this invention and should not limit this invention to the specific embodiments listed. Without departing from the technical solutions described in this invention, any equivalent substitutions, equivalent modifications, equivalent combinations, order adjustments, module corresponding substitutions, equivalent changes in data names, equivalent inheritance of execution subjects, or equivalent changes in carrier forms that can be made by those skilled in the art should fall within the scope of protection of this patent, but should not be extended to the protected subject matter not described in the application documents, nor should the substantive correspondence of the technical objects be changed through changes in names or reference numerals.

Claims

1. A method for virtual fitting of 3D clothing using multi-fabric physical simulation, characterized in that, The method includes: Collect fabric characteristic data, calibration data, and texture data; standardize the fabric characteristic data; and store the fabric characteristic data, calibration data, and texture data according to the fabric identifier to construct a fabric parameter database. Acquire clothing mesh, target fabric identifier, human body model and pose sequence; read target fabric data from the fabric parameter database according to the target fabric identifier; extract target texture data from the target fabric data; and establish an initial texture mapping between the target texture data and the clothing mesh. The current simulation parameters are generated based on the target fabric data. The calibration simulation is performed based on the calibration data and the deformation deviation is calculated. When the deformation deviation exceeds the calibration range, the current simulation parameters are updated and the calibration simulation is repeated until the deformation deviation enters the calibration range. The current simulation parameters are then determined as the target simulation parameters. The human body model is driven by the posture sequence. The target convergence condition and solution configuration are determined based on the changes in adjacent posture frames and the target simulation parameters. Physical simulation is performed based on the human body model, the clothing mesh, the target simulation parameters and the solution configuration to obtain the clothing simulation state. When the clothing simulation state does not meet the target convergence condition, the deviation region is located according to the state deviation, the solution configuration corresponding to the deviation region is updated, and a local physical simulation is performed before re-judging. When the clothing simulation state meets the target convergence condition, the initial texture mapping is updated according to the clothing simulation state and three-dimensional rendering is performed to output a three-dimensional fitting screen.

2. The multi-fabric physical simulation 3D clothing virtual try-on method according to claim 1, characterized in that, The fabric characteristic data are standardized into fabric parameters after being standardized in terms of dimensions and verified for data validity. Each data record in the fabric parameter database is indexed by the fabric identifier, and the standardized fabric parameters, the calibration data, and the texture data are associated and written into the same data record.

3. The multi-fabric physical simulation 3D clothing virtual try-on method according to claim 1, characterized in that, Retrieve data records indexed by the target fabric identifier according to the target fabric identifier, and determine the retrieved data records as the target fabric data; Based on the topological relationship between the grid vertices and grid faces in the clothing mesh, a set of texture coordinates corresponding to the grid vertices is generated. The arrangement direction of the set of texture coordinates is corrected according to the local direction of the grid faces. The target texture data is then mapped to the clothing mesh according to the corrected set of texture coordinates to form the initial texture mapping.

4. The multi-fabric physical simulation 3D clothing virtual try-on method according to claim 1, characterized in that, Target calibration data is extracted from the target fabric data, and the target calibration data characterizes the measured deformation state of the target fabric corresponding to the target fabric data under boundary conditions. Based on the clothing mesh and the current simulation parameters, the calibration simulation is performed according to the boundary conditions to obtain the virtual deformation state; The deformation deviation is generated by comparing the corresponding deformation state components in the virtual deformation state with those in the measured deformation state.

5. The multi-fabric physical simulation 3D clothing virtual try-on method according to claim 4, characterized in that, When any of the deformation state components in the deformation deviation exceeds the calibration range, the deviation type is determined according to the deformation state component that exceeds the calibration range, the parameter item corresponding to the deviation type in the current simulation parameters is updated, and the calibration simulation is re-executed according to the boundary conditions. When all the deformation state components fall within the calibration range, the calibration simulation is terminated, and the current simulation parameters are determined as the target simulation parameters.

6. The multi-fabric physical simulation 3D clothing virtual try-on method according to claim 1, characterized in that, The posture change state is determined based on the human joint motion and human surface displacement corresponding to adjacent posture frames in the posture sequence. Based on the degree of posture change characterized by the posture change state and the physical properties of the fabric characterized by the target simulation parameters, the calculation step size and solution rounds of the clothing mesh are determined to form the solution configuration; The convergence condition of the target is determined based on the deformation constraints defined by the target simulation parameters.

7. The multi-fabric physical simulation 3D clothing virtual try-on method according to claim 1, characterized in that, Calculate the constraint residuals of each grid region in the clothing simulation state, determine the state deviation based on the difference between the constraint residuals and the convergence range defined by the target convergence condition, and determine the grid region where the constraint residuals exceed the convergence range as the deviation region; The solver configuration is adjusted according to the state deviation corresponding to the deviation region. Based on the adjusted solver configuration, the local physical simulation is performed on the deviation region to update the clothing simulation state and re-determine whether the clothing simulation state meets the target convergence condition.

8. The multi-fabric physical simulation 3D clothing virtual try-on method according to claim 1, characterized in that, When the clothing simulation state satisfies the target convergence condition, the local deformation relationship is determined based on the change of vertex position of the mesh patch in the clothing mesh before and after the physical simulation, and the initial texture mapping is updated based on the local deformation relationship to form the updated texture mapping; The 3D rendering is performed based on the human body model, the clothing mesh processed by the physical simulation, and the updated texture mapping, and the 3D fitting screen is output.

9. A multi-fabric physical simulation 3D clothing virtual try-on system, the system being used to implement the multi-fabric physical simulation 3D clothing virtual try-on method as described in any one of claims 1-8, characterized in that, The system includes: The parameter database module collects fabric characteristic data, calibration data, and texture data, standardizes the fabric characteristic data, and stores the fabric characteristic data, calibration data, and texture data according to the fabric identifier to construct a fabric parameter database. The data mapping module acquires clothing mesh, target fabric identifier, human body model and pose sequence, reads target fabric data from the fabric parameter database according to the target fabric identifier, extracts target texture data from the target fabric data, and establishes an initial texture mapping between the target texture data and the clothing mesh. The parameter calibration module generates current simulation parameters based on the target fabric data, performs calibration simulation and calculates deformation deviation based on the calibration data, and updates the current simulation parameters and repeats the calibration simulation when the deformation deviation exceeds the calibration range until the deformation deviation enters the calibration range, and then determines the current simulation parameters as the target simulation parameters. The physical simulation module drives the human body model based on the posture sequence, determines the target convergence condition and solution configuration based on the changes in adjacent posture frames and the target simulation parameters, and performs physical simulation based on the human body model, the clothing mesh, the target simulation parameters and the solution configuration to obtain the clothing simulation state; The feedback rendering module, when the clothing simulation state does not meet the target convergence condition, locates the deviation region according to the state deviation, updates the solution configuration corresponding to the deviation region, and performs local physical simulation and then re-judges. When the clothing simulation state meets the target convergence condition, it updates the initial texture mapping according to the clothing simulation state and performs 3D rendering to output a 3D fitting screen.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the multi-fabric physical simulation 3D clothing virtual fitting system as described in claim 9.

Images